Datarails aims for ‘grounded’ ChatGPT-like FP&A tool

Dive Brief:

Software provider Datarails is testing new artificial intelligence capabilities in a financial planning and analysis tool for CFOs, company executives told CFO Dive.
The platform, known as “FP&A Genius,” is currently in beta form and will be fully rolled out in the third quarter, Datarails CEO Didi Gurfinkel said in an interview on Monday.
The goal is to provide CFOs with an AI-powered conversational tool in the style of the popular ChatGPT service that was developed by Microsoft-backed OpenAI, but without some of the same accuracy and data privacy challenges, according to Gurfinkel. “We cannot release a product that will give wrong answers to the CFO,” he said.

Dive Insight:

Datarails is one of a growing list of companies that are moving quickly to adopt ChatGPT-style tools for finance departments.

ChatGPT is an AI-driven natural language processing tool capable of interacting in a conversational way and producing responses to questions across a number of subject areas.

“I think ChatGPT has taken the world by storm and has awakened an NLP free-for-all,” Sloane Kolt, head of Datarails Labs, said in the same interview.

Despite its popularity, ChatGPT has also prompted concerns. One criticism is that it asserts incorrect facts at times.

Datarails is using the current beta testing period for FP&A Genius to ensure the platform will not engage in the kind of “hallucinations” that are associated with ChatGPT, Kolt said.

“Finance is a complex topic,” she said. “We want to make sure we’re building something that’s grounded. We have a company that has a tremendous amount of ex-finance people working here, and we understand the stakes.”

ChatGPT, which is available on the public internet, has also been associated with data privacy and security concerns.

Gurfinkel said FP&A Genius combines the company’s own proprietary analytics tools with open source generative AI and NLP technologies. “Everything is run in our secure cloud environment,” he said.

New York-based Datarails was founded in 2015 and has 160 employees, with nearly 1,000 clients, primarily in the U.S. and Canada.

Earlier this month, the company announced that it raised $50 million in Series B funding — its third investment round in the past 12 months. To date, the company has raised a total of $105 million in funding, Datarails spokesperson Jonathan Marciano told CFO Dive.

On its website, Datarails describes FP&A Genius as an AI-powered chat service that provides “answers based on complete and consolidated finance data from across a company.” Examples of how finance leaders can use the service include asking it to compare an organization’s revenue in 2022 versus 2021, or to determine what a company’s bottom line would be if inflation were twice the current rate, according to Marciano.

In a related move, San Francisco-based tech startup Klarity Intelligence announced on March 15 that it began using GPT-4, the latest model of OpenAI’s technology, in its platform, which is designed to enable users to more quickly and easily perform functions such as validating billing and revenue data.

Klarity’s upgraded platform is currently available as a demo as the company works to address data security issues and other potential concerns before rolling out an enterprise-ready version by sometime in April, Nischal Nadhamuni, the startup’s chief technology officer and co-founder, previously told CFO Dive.

OpenAI said in a March 14 press release that GPT-4 “poses similar risks as previous models, such as generating harmful advice, buggy code, or inaccurate information.”

Meanwhile, finance tech startup Brex, also based in San Francisco, has said it’s incorporating new ChatGPT-style features into its “Empower” platform, a spending management tool. The new features “will incorporate data from millions of transactions in order to provide relevant comparables that help benchmark performance and spend, all while maintaining and preserving strict privacy,” according to a March 7 press release.